A Supervised Classifier for Police Reports at the State of Para, Brazil

Abstract

This paper describes the development of a supervised classifier constructed upon knowledge extracted from police report public databases, in the years between 2019 and 2021 in the state of Pará, Brazil. The classifier achieved an accuracy of approximately 78% for the prediction of 463 unique labels related to public safety. The resulting model can be used to improve the statistical processes of criminal analysts, both in quantitative and qualitative terms.